Results

cesm2.ssp245

diff_of_means ratio_of_sd monthly_amplitude_ratio_of_means ks_mean_on_coarse_res_with_extremes qqplot_mae acf_mae extremogram_mae
lstm.cesm2.ssp245 -3.44% 0.816 0.811 0.187 4.681 0.025 0.022
cnn.cesm2.ssp245 -3.62% 0.823 0.875 0.179 5.755 0.029 0.027
xgboost.cesm2.ssp245 -7.04% 0.791 0.795 0.227 5.990 0.031 0.021
nv.cesm2.ssp245 -21.42% 0.830 0.882 0.272 10.557 0.060 0.019

Time series of the first days

Distribution of daily values by month

QQ Plot

Distribution of the undownscaled value on days with estimated extremes values.

On the x-axis we have the daily mean (standardized). It says Undownscaled value, but is the daily mean after the downscaling. A good idea is to plot the original undownscaled value.

The purpose of this plot is to illustrate the distribution of P(undownscaled value | we predicted an extreme). This is useful because it reveals how much information we can recover concerning extreme events. If the distribution is skewed to the right, it suggests that we’re predicting extreme values only when extreme values have already occurred. Conversely, if the lower tail of the distribution resembles the reanalysis data, it indicates that we can capture short-duration extremes (e.g., brief periods of heavy rainfall, such as an intense downpour lasting an hour before stopping).

Autocorrelogram

Extremogram

cesm2.ssp370

diff_of_means ratio_of_sd monthly_amplitude_ratio_of_means ks_mean_on_coarse_res_with_extremes qqplot_mae acf_mae extremogram_mae
lstm.cesm2.ssp370 -3.05% 0.813 0.803 0.109 4.818 0.029 0.024
cnn.cesm2.ssp370 -3.51% 0.800 0.828 0.174 6.115 0.031 0.022
xgboost.cesm2.ssp370 -9.62% 0.774 0.773 0.176 6.883 0.033 0.022
nv.cesm2.ssp370 -24.55% 0.841 0.885 0.195 12.067 0.063 0.026

Time series of the first days

Distribution of daily values by month

QQ Plot

Distribution of the undownscaled value on days with estimated extremes values.

On the x-axis we have the daily mean (standardized). It says Undownscaled value, but is the daily mean after the downscaling. A good idea is to plot the original undownscaled value.

The purpose of this plot is to illustrate the distribution of P(undownscaled value | we predicted an extreme). This is useful because it reveals how much information we can recover concerning extreme events. If the distribution is skewed to the right, it suggests that we’re predicting extreme values only when extreme values have already occurred. Conversely, if the lower tail of the distribution resembles the reanalysis data, it indicates that we can capture short-duration extremes (e.g., brief periods of heavy rainfall, such as an intense downpour lasting an hour before stopping).

Autocorrelogram

Extremogram

cesm2.ssp585

diff_of_means ratio_of_sd monthly_amplitude_ratio_of_means ks_mean_on_coarse_res_with_extremes qqplot_mae acf_mae extremogram_mae
cnn.cesm2.ssp585 -2.03% 0.781 0.836 0.160 6.446 0.027 0.026
lstm.cesm2.ssp585 -2.65% 0.799 0.810 0.165 5.157 0.025 0.029
xgboost.cesm2.ssp585 -9.44% 0.778 0.780 0.224 6.753 0.031 0.020
nv.cesm2.ssp585 -25.09% 0.828 0.909 0.107 12.330 0.054 0.022

Time series of the first days

Distribution of daily values by month

QQ Plot

Distribution of the undownscaled value on days with estimated extremes values.

On the x-axis we have the daily mean (standardized). It says Undownscaled value, but is the daily mean after the downscaling. A good idea is to plot the original undownscaled value.

The purpose of this plot is to illustrate the distribution of P(undownscaled value | we predicted an extreme). This is useful because it reveals how much information we can recover concerning extreme events. If the distribution is skewed to the right, it suggests that we’re predicting extreme values only when extreme values have already occurred. Conversely, if the lower tail of the distribution resembles the reanalysis data, it indicates that we can capture short-duration extremes (e.g., brief periods of heavy rainfall, such as an intense downpour lasting an hour before stopping).

Autocorrelogram

Extremogram

ec_earth3.ssp434

diff_of_means ratio_of_sd monthly_amplitude_ratio_of_means ks_mean_on_coarse_res_with_extremes qqplot_mae acf_mae extremogram_mae
lstm.ec_earth3.ssp434 -6.33% 0.874 0.856 0.219 3.699 0.023 0.026
cnn.ec_earth3.ssp434 -6.88% 0.802 0.864 0.232 5.941 0.030 0.019
xgboost.ec_earth3.ssp434 -23.42% 0.759 0.758 0.423 11.728 0.078 0.036
nv.ec_earth3.ssp434 -32.87% 0.783 0.846 0.292 16.156 0.070 0.024

Time series of the first days

Distribution of daily values by month

QQ Plot

Distribution of the undownscaled value on days with estimated extremes values.

On the x-axis we have the daily mean (standardized). It says Undownscaled value, but is the daily mean after the downscaling. A good idea is to plot the original undownscaled value.

The purpose of this plot is to illustrate the distribution of P(undownscaled value | we predicted an extreme). This is useful because it reveals how much information we can recover concerning extreme events. If the distribution is skewed to the right, it suggests that we’re predicting extreme values only when extreme values have already occurred. Conversely, if the lower tail of the distribution resembles the reanalysis data, it indicates that we can capture short-duration extremes (e.g., brief periods of heavy rainfall, such as an intense downpour lasting an hour before stopping).

Autocorrelogram

Extremogram

mri_esm2_0.ssp245

diff_of_means ratio_of_sd monthly_amplitude_ratio_of_means ks_mean_on_coarse_res_with_extremes qqplot_mae acf_mae extremogram_mae
nv.mri_esm2_0.ssp245 1.43% 0.790 0.847 0.232 7.561 0.132 0.019
cnn.mri_esm2_0.ssp245 -2.27% 0.723 0.785 0.126 7.971 0.038 0.024
lstm.mri_esm2_0.ssp245 -3.44% 0.780 0.801 0.112 5.722 0.035 0.021
xgboost.mri_esm2_0.ssp245 13.19% 0.830 0.812 0.306 7.938 0.042 0.028

Time series of the first days

Distribution of daily values by month

QQ Plot

Distribution of the undownscaled value on days with estimated extremes values.

On the x-axis we have the daily mean (standardized). It says Undownscaled value, but is the daily mean after the downscaling. A good idea is to plot the original undownscaled value.

The purpose of this plot is to illustrate the distribution of P(undownscaled value | we predicted an extreme). This is useful because it reveals how much information we can recover concerning extreme events. If the distribution is skewed to the right, it suggests that we’re predicting extreme values only when extreme values have already occurred. Conversely, if the lower tail of the distribution resembles the reanalysis data, it indicates that we can capture short-duration extremes (e.g., brief periods of heavy rainfall, such as an intense downpour lasting an hour before stopping).

Autocorrelogram

Extremogram

mri_esm2_0.ssp370

diff_of_means ratio_of_sd monthly_amplitude_ratio_of_means ks_mean_on_coarse_res_with_extremes qqplot_mae acf_mae extremogram_mae
cnn.mri_esm2_0.ssp370 -5.98% 0.738 0.793 0.187 7.585 0.037 0.023
xgboost.mri_esm2_0.ssp370 6.57% 0.851 0.820 0.321 5.274 0.040 0.023
nv.mri_esm2_0.ssp370 -7.00% 0.821 0.883 0.181 6.921 0.117 0.021
lstm.mri_esm2_0.ssp370 -7.66% 0.782 0.798 0.264 5.987 0.023 0.024

Time series of the first days

Distribution of daily values by month

QQ Plot

Distribution of the undownscaled value on days with estimated extremes values.

On the x-axis we have the daily mean (standardized). It says Undownscaled value, but is the daily mean after the downscaling. A good idea is to plot the original undownscaled value.

The purpose of this plot is to illustrate the distribution of P(undownscaled value | we predicted an extreme). This is useful because it reveals how much information we can recover concerning extreme events. If the distribution is skewed to the right, it suggests that we’re predicting extreme values only when extreme values have already occurred. Conversely, if the lower tail of the distribution resembles the reanalysis data, it indicates that we can capture short-duration extremes (e.g., brief periods of heavy rainfall, such as an intense downpour lasting an hour before stopping).

Autocorrelogram

Extremogram

mri_esm2_0.ssp434

diff_of_means ratio_of_sd monthly_amplitude_ratio_of_means ks_mean_on_coarse_res_with_extremes qqplot_mae acf_mae extremogram_mae
nv.mri_esm2_0.ssp434 -3.17% 0.816 0.850 0.198 6.919 0.155 0.025
cnn.mri_esm2_0.ssp434 -4.64% 0.739 0.799 0.075 7.756 0.058 0.034
lstm.mri_esm2_0.ssp434 -6.25% 0.776 0.784 0.182 5.980 0.041 0.020
xgboost.mri_esm2_0.ssp434 9.10% 0.829 0.803 0.250 6.402 0.056 0.042

Time series of the first days

Distribution of daily values by month

QQ Plot

Distribution of the undownscaled value on days with estimated extremes values.

On the x-axis we have the daily mean (standardized). It says Undownscaled value, but is the daily mean after the downscaling. A good idea is to plot the original undownscaled value.

The purpose of this plot is to illustrate the distribution of P(undownscaled value | we predicted an extreme). This is useful because it reveals how much information we can recover concerning extreme events. If the distribution is skewed to the right, it suggests that we’re predicting extreme values only when extreme values have already occurred. Conversely, if the lower tail of the distribution resembles the reanalysis data, it indicates that we can capture short-duration extremes (e.g., brief periods of heavy rainfall, such as an intense downpour lasting an hour before stopping).

Autocorrelogram

Extremogram